22 October 2002

Appendix C: Accessing code point boundaries

Mark Davis, IBM
Lauren Wood, SoftQuad Software Inc.

Table of contents

C.1: Introduction

This appendix is an informative, not a normative, part of the Level 2 DOM specification.

Characters are represented in Unicode by numbers called code points (also called scalar values). These numbers can range from 0 up to 1,114,111 = 10FFFF16 (although some of these values are illegal). Each code point can be directly encoded with a 32-bit code unit. This encoding is termed UCS-4 (or UTF-32). The DOM specification, however, uses UTF-16, in which the most frequent characters (which have values less than FFFF16) are represented by a single 16-bit code unit, while characters above FFFF16 use a special pair of code units called a surrogate pair. For more information, see [Unicode 2.0] or the Unicode Web site.

While indexing by code points as opposed to code units is not common in programs, some specifications such as XPath (and therefore XSLT and XPointer) use code point indices. For interfacing with such formats it is recommended that the programming language provide string processing methods for converting code point indices to code unit indices and back. Some languages do not provide these functions natively; for these it is recommended that the native String type that is bound to DOMString be extended to enable this conversion. An example of how such an API might look is supplied below.

Note: Since these methods are supplied as an illustrative example of the type of functionality that is required, the names of the methods, exceptions, and interface may differ from those given here.

C.2: Methods

Interface StringExtend

Extensions to a language's native String class or interface


IDL Definition
interface StringExtend {
  int                findOffset16(in int offset32)
                                        raises(StringIndexOutOfBoundsException);
  int                findOffset32(in int offset16)
                                        raises(StringIndexOutOfBoundsException);
};

Methods
findOffset16
Returns the UTF-16 offset that corresponds to a UTF-32 offset. Used for random access.

Note: You can always round-trip from a UTF-32 offset to a UTF-16 offset and back. You can round-trip from a UTF-16 offset to a UTF-32 offset and back if and only if the offset16 is not in the middle of a surrogate pair. Unmatched surrogates count as a single UTF-16 value.

Parameters
offset32 of type int
UTF-32 offset.
Return Value

int

UTF-16 offset

Exceptions

StringIndexOutOfBoundsException

if offset32 is out of bounds.

findOffset32
Returns the UTF-32 offset corresponding to a UTF-16 offset. Used for random access. To find the UTF-32 length of a string, use:
len32 = findOffset32(source, source.length());

Note: If the UTF-16 offset is into the middle of a surrogate pair, then the UTF-32 offset of the end of the pair is returned; that is, the index of the char after the end of the pair. You can always round-trip from a UTF-32 offset to a UTF-16 offset and back. You can round-trip from a UTF-16 offset to a UTF-32 offset and back if and only if the offset16 is not in the middle of a surrogate pair. Unmatched surrogates count as a single UTF-16 value.

Parameters
offset16 of type int
UTF-16 offset
Return Value

int

UTF-32 offset

Exceptions

StringIndexOutOfBoundsException

if offset16 is out of bounds.